Fault detection for appliances based on energy consumption data

Abstract

A method of detecting a fault in a device is described. The method comprises receiving an energy consumption signal for the device, the energy consumption signal indicating a varying energy consumption of the device over time during operation of the device. A basis function is selected, the basis function defining a signal pattern associated with a known fault type. A correlation between the basis function and the energy consumption signal is evaluated. Presence of an operating fault in the device is determined in dependence on the correlation, and an output is generated based on the determination.

The invention claimed is: 
     
       1. A computer-implemented method of detecting a fault in a device, comprising:
 receiving an energy consumption signal for the device, the energy consumption signal indicating a varying energy consumption of the device over time during operation of the device; 
 selecting at least one basis function from a database storing information defining a plurality of basis functions, wherein each basis function defines a signal pattern associated with a respective known fault type, each basis function having a shape which matches the shape of an expected consumption signal when the respective fault type is present; 
 repeatedly determining a correlation between the at least one basis function and the energy consumption signal, while varying one or more parameters of the basis function, wherein determining the correlation comprises computing correlation values between a plurality of portions of the energy consumption signal and the basis function for each of a plurality of values of the one or more parameters; 
 determining presence of an operating fault in the device in dependence on the determined correlation values; and 
 generating an output based on the determination, wherein generating the output comprises one or more of: 
 transmitting an electronic message to a user device indicating at least one of: presence of the fault, absence of the fault, and a detected fault type; and 
 transmitting a control signal to the device or a control system associated with the device to change at least one of: an activation state of the device, an operating state of the device, and an operating configuration of the device. 
 
     
     
       2. A method according to  claim 1 , wherein the energy consumption signal comprises a time series of energy consumption values extending over a given time period, the evaluating step comprising evaluating a correlation between the basis function and the energy consumption signal at each of a plurality of times or time windows of the energy consumption signal. 
     
     
       3. A method according to  claim 2 , comprising computing a time series of intensity values, each intensity value indicative of a strength of correlation of the basis function to the energy consumption signal at a respective time or time window, wherein computing a time series of intensity values comprises computing a respective intensity value for each of a plurality of time instants based on a correlation of the basis function when centred at a respective time instant, wherein a width of the basis function defines a time window of the energy consumption signal over which correlation is evaluated. 
     
     
       4. A method according to  claim 3 , comprising:
 analysing the intensity values to identify one or more intensity values meeting a detection criterion, wherein the detection criterion is based on comparison of the intensity values to a comparison threshold, the detection criterion comprising at least one intensity value exceeding the comparison threshold; and 
 in response to identifying one or more intensity values meeting the detection criterion, detecting an operating fault of the known fault type, or 
 in response to identifying no intensity values meeting the detection criterion, identifying that the fault is not present. 
 
     
     
       5. A method according to  claim 1 , wherein repeating the evaluation while varying one or more parameters of the basis function comprises varying each of the one or more parameters over a predefined set of parameter values or over a parameter range. 
     
     
       6. A method according to  claim 1 , wherein the one or more parameters comprise one or more of: a width parameter indicating a width of the basis function and/or signal window over which correlation is performed; and a shape parameter controlling a shape of the basis function. 
     
     
       7. A method according to  claim 1 , comprising at least one of:
 computing at least one of: a representative correlation intensity, an average correlation intensity, and a maximum correlation intensity, over a range of variation of one or more basis function parameters; and 
 determining, for each of a plurality of values of a first parameter at least one of: a representative correlation intensity, an average correlation intensity, and a maximum correlation intensity, over a range of variation of a second parameter, wherein the first parameter is a shape parameter and the second parameter is a width parameter. 
 
     
     
       8. A method according to  claim 1 , comprising repeating the evaluation for each of a plurality of fault types, using one or more respective predetermined basis functions defined for each respective fault type, to detect presence or absence of each of the fault types, wherein the outputting step comprises, in response to detection of a given fault type based on correlation of a basis function associated with the given fault with the energy consumption signal, outputting an indication of the detected fault type. 
     
     
       9. A method according to  claim 1 , wherein evaluating a correlation between the at least one basis function and the energy consumption signal comprises computing one or more measures of correlation between the basis function and the energy consumption signal, and comparing the measure(s) against one or more comparison values, the comparison value(s) having been established based on energy consumption of a non-faulty device. 
     
     
       10. A method according to  claim 1 , further comprising establishing comparison values for use in performing the fault detection based on a reference energy consumption signal of a device operated in the absence of a fault. 
     
     
       11. A method according to  claim 10 , comprising, for a given fault type, or for each given fault type of a plurality of fault types:
 evaluating a basis function associated with the given fault type against the reference energy consumption signal, by determining one or more correlation intensities indicating strength of correlation between the basis function and the reference energy consumption signal; and 
 determining a comparison value for the fault type based on the evaluation. 
 
     
     
       12. A method according to  claim 11 , wherein determining one or more correlation intensities comprises determining a time series of correlation intensities indicating strength of correlation between the basis function and the reference energy consumption signal at each of a plurality of times or time windows of the energy consumption signal, and comprising determining a comparison value based on the time series of correlation intensities, by selecting or computing a representative or maximum correlation intensity from the time series of correlation intensities. 
     
     
       13. A method according to  claim 11 , wherein evaluating a basis function against the reference energy consumption signal comprises varying one or more basis function parameters over predetermined parameter ranges to determine representative correlation intensities. 
     
     
       14. A method according to  claim 11 , wherein detection of a fault is performed based on comparing the comparison value obtained using the reference energy consumption signal to correlation intensities determined for the energy consumption signal. 
     
     
       15. A method according to  claim 1 , comprising receiving user input for configuring, for one or more fault types, one or more of:
 at least one basis function suitable for detecting the fault type; 
 one or more ranges for one or more parameters of a basis function; 
 one or more limits for one or more parameters of a basis function; and 
 one or more detection criteria for the fault type; 
 and wherein the evaluation is performed in accordance with the configured information. 
 
     
     
       16. A method according to  claim 1 , wherein the energy consumption signal and/or reference energy consumption signal comprise electricity consumption signals, measured using an electricity consumption measurement device. 
     
     
       17. A non-transitory computer-readable medium comprising software code adapted, when executed on a processing device, to perform a method of detecting a fault in a device, comprising:
 receiving an energy consumption signal for the device, the energy consumption signal indicating a varying energy consumption of the device over time during operation of the device; 
 selecting at least one basis function from a database storing information defining a plurality of basis functions, wherein each basis function defines a signal pattern associated with a respective known fault type; each basis function having a shape which matches the shape of an expected consumption signal when the respective fault type is present; 
 repeatedly determining a correlation between the at least one basis function and the energy consumption signal while varying one or more parameters of the basis function, wherein determining the correlation comprises computing correlation values between a plurality of portions of the energy consumption signal and the basis function for each of a plurality of values of the one or more parameters; 
 determining presence of an operating fault in the device in dependence on the determined correlation values; and 
 generating an output based on the determination, wherein generating the output comprises one or more of: 
 transmitting an electronic message to a user device indicating at least one of: presence of the fault, absence of the fault, and a detected fault type; and 
 transmitting a control signal to the device or a control system associated with the device to change at least one of: an activation state of the device, an operating state of the device, and an operating configuration of the device. 
 
     
     
       18. A computer-implemented method of detecting a fault in a device, comprising:
 receiving an energy consumption signal for the device, the energy consumption signal indicating a varying energy consumption of the device over time during operation of the device; 
 selecting a basis function from a database storing information defining a plurality of basis functions, each basis function defining a signal pattern associated with a respective known fault type, each basis function having a shape which matches the shape of an expected consumption signal when the respective fault type is present; wherein the selected basis function comprises a windowed function and is parameterized by at least one parameter, the at least one parameter comprising one or more of: a width parameter indicating a width of the basis function and/or signal window over which correlation is performed; and a shape parameter controlling the shape of the basis function; 
 evaluating a correlation between the basis function and the energy consumption signal, wherein evaluating the correlation comprises computing correlation values between a plurality of portions of the energy consumption signal and the basis function for one or more values of the at least one parameter; 
 determining presence of an operating fault in the device in dependence on the correlation values; and 
 generating an output based on the determination, wherein generating the output comprises one or more of: 
 transmitting an electronic message to a user device indicating at least one of: presence of the fault, absence of the fault, and a detected fault type; and 
 transmitting a control signal to the device or a control system associated with the device to change at least one of: an activation state of the device, an operating state of the device, and an operating configuration of the device. 
 
     
     
       19. A method according to  claim 18 , wherein evaluating the correlation comprises repeating the evaluation while varying one or more of the parameters of the basis function, each parameter varied over a respective predefined set of parameter values.